DiffETM: Diffusion Process Enhanced Embedded Topic Model

TL;DR

DiffETM introduces a diffusion process into embedded topic modeling to better capture complex document-topic distributions, improving performance while maintaining ease of optimization.

Contribution

It proposes a novel diffusion-enhanced approach for embedded topic models, addressing the oversimplification of the logistic normal assumption.

Findings

Improved topic modeling accuracy on benchmark datasets

Effective in capturing complex document-topic distributions

Maintains ease of optimization

Abstract

The embedded topic model (ETM) is a widely used approach that assumes the sampled document-topic distribution conforms to the logistic normal distribution for easier optimization. However, this assumption oversimplifies the real document-topic distribution, limiting the model's performance. In response, we propose a novel method that introduces the diffusion process into the sampling process of document-topic distribution to overcome this limitation and maintain an easy optimization process. We validate our method through extensive experiments on two mainstream datasets, proving its effectiveness in improving topic modeling performance.